Legislación comparada sobre libertad de información

In general, the pharmacophore models had certain characteristic traits that were common across all protein targets investigated in this study. Pharmacophore models from NMR structures had fewer pharmacophore elements which were greater in size, as compared to their crystal-pharmacophore counterparts.

2.4.2.1 Comparison of the Src SH2 pharmacophore models

Src SH2 is an important component in the auto regulation of its kinase domain; upon phosphorylation the C-terminus of the protein binds to the SH2 domain and results in the distortion of the kinase active site (82). The SH2 domain of Src binds with high affinity to phosphorylated peptides and recognizes the peptide sequence pYEEI with high affinity (83). The phosphotyrosine moiety of peptide ligands bind to the pY site, and the pY+1, pY+2, pY+3 sub-sites that determine specificity lie C terminal to this phosphotyrosine binding site (Figure 2-3A, Figure 2-3B). Pharmacophore models from both crystal and NMR structures reproduced key essential features in the different sub-sites required for binding substrates. As shown in Figure 2-3B, the NMR pharmacophore model for Src SH2 had six pharmacophore elements compared to the ten pharmacophore elements in the crystal model (Figure 2-3A). The pY site in the

Protein Crystal Structures Binding Site RMSD

NMR Ensemble Binding Site RMSD

Grb2 0.33 - 1.71 Å 1.02 - 2.08 Å

Src SH2 0.29 - 1.18 Å 1.00 - 1.79 Å

FKBP12 0.64 - 2.57 Å 1.58 - 3.95 Å

PPAR-γ 1.37 - 2.03 Å 1.49 - 3.21 Å

34

crystal pharmacophore model had two extra elements, a donor and a doneptor that were absent in the NMR model. Additionally, an extra hydrophobic element in the pY+2 and an aromatic element in the pY+3 pockets were located in the crystal pharmacophore model. The elements in the NMR pharmacophore model represented a subset of those seen in the crystal pharmacophore model. Interestingly, while the exact location of the phosphotyrosine moiety was not mapped in the NMR and crystal pharmacophore models, a doneptor element in close proximity was identified whose location appeared shifted between the two pharmacophore models. The lone bonding element in the pY+3 pocket changed from a hydrogen-bond donor in the crystal pharmacophore model to a hydrogen-hydrogen-bond acceptor in the NMR pharmacophore model. This resulted from differing positions of a tyrosine residue lining the pY+3 pocket. Overlaying the crystal-structure ligands with the pharmacophore models as shown in Figure 2-3C emphasizes the observation that most elements from the NMR model overlap with the ligands in contrast to the crystal model where many elements failed to do so. This is particularly surprising because the ligands in Figure 2-3D are from the crystal structures, not the NMR model.

35

Figure 2-3. MPS pharmacophores are shown with 1× RMSD radii, which indicates tighter or looser position constraints. Pharmacophore models for all the protein targets are color coded to represent different interactions: Red – Donor, Blue – Acceptor, Purple – Doneptor, Green – Aromatic, and Cyan – Hydrophobic. A) The MPS pharmacophore model for Src SH2 derived from X-ray structures. B) The MPS pharmacophore model for Src SH2 derived from the NMR ensemble. C) The ligands from X-ray structures overlaid on top of the Src SH2 X-ray model. D) The ligands from X-ray structures overlaid on top of the Src SH2 NMR model.

2.4.2.2 Comparison of the Grb2 SH2 pharmacophore models

Grb2 is an adaptor protein and consists of two SH3 domains and one SH2 domain. The SH2 domain of Grb2 binds to phosphorylated peptides of the general sequence pYXNX and adopts a

36

similar fold seen in other SH2 domains such as Src SH2. Sub-sites in the protein active site that accommodate ligands follow a similar nomenclature as noted above for Src SH2 and are named pY, pY+1, pY+2 (show in Figure 2-4A and Figure 2-4B). A key difference between the Grb2 and Src SH2 domains is Trp 121 in Grb2 which is part of the specificity determining EF loop (according to the naming convention described in Eck et al. (84)) that blocks the large pY+3 pocket seen in Src SH2. As a result, phosphotyrosine peptides that bind to Grb2 SH2 domain adopt a beta turn instead of binding in an extended conformation occupying the pY+3 sub-site in Src SH2 (85). MPS pharmacophore models in this study were obtained for the SH2 domain of Grb2. Pharmacophore models reproduced key features of the active site which included the phosphotyrosine binding location in the pY subsite and essential interactions seen across all ligands in the pY+1 pocket (shown in Figure 2-4A and Figure 2-4B). A hydrophobic element that overlapped the benzene ring of the phosphotyrosine residue was seen in the pY subsite of both pharmacophore models (Figure 2-4C and Figure 2-4D). The pY subsite of the crystal pharmacophore model displayed an additional acceptor element not found in the NMR model.

As seen in Figure 2-4C, this acceptor element overlaps with the carbonyl group of the amide bond linking the phosphotyrosine residue to the residue preceding it. Three pharmacophore elements in the pY+1 pocket appear in similar locations in the crystal and NMR models. The only difference between the two pharmacophore models in the pY+1 pocket was an additional doneptor element in the NMR model. The pY+2 sub site followed a similar trend and had more elements in the crystal pharmacophore model. A doneptor and hydrophobic element consistent with a key interaction is seen in both NMR and Crystal pharmacophore model (see Figure 2-4C and Figure 2-4D).

37

Figure 2-4. Coloring and radii of the pharmacophore elements are the same as in Figure 2-3. A) The MPS pharmacophore model for GRB2 SH2 derived from X-ray structures. B) The MPS pharmacophore model for GRB2 SH2 derived from the NMR ensemble. C) The ligands from X-ray structures overlaid on top of the GRB2 SH2 MPS X-X-ray pharmacophore model. D) The ligands from X-ray structures overlaid on top of the GRB2 SH2 MPS NMR pharmacophore model. In Figure 2-4A and Figure 2-4B, tryptophan 121 is rendered as sticks between the pY+1 and pY+2 surfaces.

38

2.4.2.3 Comparison of the FKBP12 pharmacophore models

FKBP12 is a peptidyl prolyl cis/trans-isomerase that catalyzes the isomerization of proline amide bonds in proteins and peptides, and it is known to act as an immunosuppressant in complex with FK506 or Rapamycin (86, 87). The proline ring of a substrate sits in the center of the active site, which is a hydrophobic pocket lined with aromatic and hydrophobic residues with a tryptophan residue forming the base of the pocket (88, 89). Pharmacophore models from both crystal and NMR structures identify this key hydrophobic pocket in the center of the active site as illustrated in Figure 2-5A and Figure 2-5B. While the crystal pharmacophore model identifies several elements at the periphery of the active site, the absence of most of these elements from the NMR pharmacophore model is quiet apparent and can be attributed to the lack of consensus clusters in the more flexible regions of the NMR ensemble of FKBP12. An overlay of the pharmacophore models with the ligands bound to FKBP12 in the crystal structures (as seen in Figure 2-5C and Figure 2-5D) provides a more detailed understanding of the location of the elements. The only hydrogen-bonding element in the NMR pharmacophore model is a doneptor element that closely overlaps the carboxylic acid region of the proline residue in the ligands, making hydrogen-bonding interactions with the backbone of the protein (Figure 2-5D).

It is interesting to note that some of the elements that are exclusive to the crystal pharmacophore model do not overlap with any ligands from the FKBP12 protein ligand complexes. This does not appear to be the case for the NMR model where most ligands from the crystal structures overlap with all pharmacophore elements of the NMR model.

39

Figure 2-5. Coloring and radii of the pharmacophore elements are the same as in Figure 2-3. A) The MPS pharmacophore model for FKBP12 derived from X-ray structures. B) The MPS pharmacophore model for FKBP12 derived from the NMR ensemble. C) The ligands from X-ray structures overlaid on top of the FKBP12 X-ray model. D) The ligands from X-ray structures overlaid on top of the FKBP12 NMR model.

2.4.2.4 Comparison of the PPAR-γ pharmacophore models

PPAR-γ is a ligand-activated transcriptional factor. It primarily consists of a ligand-binding domain and a DNA-binding domain (90). PPAR-γ agonists bind to the ligand-binding domain and stabilize helix 12 located at the C-terminus, resulting in a conformational change to a closed form of helix 12. PPAR-γ agonists stabilize helix 12 through a network of hydrogen bonds

40

involving Tyr 473 in helix 12 and several polar residues in the vicinity (His 449, His 323, and Ser 289) typically through a carboxylic acid or thiazolidienedione moiety (91). MPS pharmacophore models in this study were obtained by flooding at the center of the active site in the ligand-binding domain. The crystal pharmacophore model displayed six elements which included a doneptor element that mapped the functional moiety of PPAR-γ agonists that stabilizes helix 12 (see Figure 2-6A). Four of the six elements in the crystal pharmacophore model were seen to overlay well with the crystal structure ligands. As a ligand-bound NMR ensemble for PPAR-γ was not available, we had to use an apo NMR ensemble for creating the pharmacophore model. The NMR pharmacophore model had three elements and the doneptor element mimicking the key functional moiety of PPAR-γ agonists was absent (see Figure 2-6). Helix 12 in the NMR ensemble (nine apo structures) sampled a wide variety of open conformations (see Figure 2-6) that did not resemble the well-ordered, hydrogen-bonding network seen in the agonist bound PPAR-γ crystal structures. Consequently, the flexibility and the absence of a doneptor element that mapped the PPAR-γ agonist functional moiety was expected. A hydrophobic element that overlapped with the agonists near helix 12 in the crystal pharmacophore model appeared shifted in the NMR model and more closely mapped the location of the Tyr 473 seen in the crystal structures. It is important to note that while the NMR pharmacophore model mapped important locations of the protein, these locations were less important for ligand binding and of more relevance for the conformational change going from the inactive form of the protein to the ligand-bound, activated form. The only element in common between the NMR and crystal pharmacophore model was a hydrophobic element located at the entrance to the active site of PPAR-γ. Interestingly, the aromatic element in the center of the active site of the crystal pharmacophore model was replaced by a donor element in the NMR pharmacophore model, presumably due to the bent nature of the helix in this region that exposes a cysteine residue (Cys 285) backbone amide in the NMR ensemble.

41

Figure 2-6. Coloring and radii of the pharmacophore elements are the same as in Figure 2-3.

MPS pharmacophore models derived from PPAR-γ X-ray and NMR ensembles are shown along with a representative protein conformation. The tyrosine residue 473 which is part of Helix 12 is shown in pink. In the X-ray ensemble, there is limited sampling of the tyrosine residue 473, which corresponds to the active form of the protein. In the NMR ensemble; this residue samples the inactive conformation of the protein. A) The MPS pharmacophore model for PPAR-γ derived from X-ray structures. B) The MPS pharmacophore model for PPAR-PPAR-γ derived from the NMR ensemble. C) The ligands from X-ray structures overlaid on top of the PPAR-γ MPS X-ray

42

pharmacophore model. D) The ligands from X-ray structures overlaid on top of the PPAR-γ MPS NMR pharmacophore model. In Figure 2-6A and Figure 2-6B, the location of the binding site is shown by rendering rosiglitazone as a stick model obtained from the PDB ID: 1ZGY (colored brown).